Loading Real Tracking Data
Minor Planet Center Astrometry
The Minor Planet Center (MPC) collects and disseminates astrometric observations for minor solar system bodies. We have implemented an interface between Tudat and the MPC’s observation database, by making use of the astroquery Python library.
At the moment, the documentation of the Python-native MPC interface is not rendering correctly on our API documentation page. For now, you can find the in-code comments of our interface on github.
Loading MPC astrometric observations into Tudat starts with creating a
BatchMPC object. Then, data can be loaded into this object by
get_observations function, which takes a list of small body identifiers and queries all astrometric data of these bodies from the MPC.
Subsequently, the data in the object can be filtered by:
Keeping only data from a given set of observatories
Removing data from a given set of observatories
Retaining only data that falls in a given time span
BatchMPC object can be used to create an
contains all remaining data, using the
to_tudat function. Note that, in the conversion to a Tudat-compatible data set,
one has the option to filter any and all space observatories (e.g. WISE, Hubble).
For instance, the following example will retrieve all data from asteroids 433 (Eros), 238 (Hypatia) and 329 (Svea), over a period of 10 years (2010-2020)
# Create MPC data container mpc_container = BatchMPC() # Load all data from given asteroids asteroid_MPC_codes = [433, 238, 329] mpc_container.get_observations(asteroid_MPC_codes) # Filter data based on time mpc_container.filter(epoch_start=datetime.datetime(2010, 1, 1), epoch_end=datetime.datetime(2020, 1, 1) # Convert data to Tudat-compatible object observation_collection = mpc_container.to_tudat(bodies, included_satellites=None)
Several examples using MPC data can be found on our page with estimation examples
Natural Satellite Data Center Astrometry
The Natural Satellite Data Center is the largest database of astrometric observations of natural satellites of solar system planets (except Earth’s moon). Unlike astrometry from the MPC, there is no existing library to extract data from this website, and there is more diversity in the layout of the files in the NSDC.
An interface between Tudat and the NSDC astrometry is under construction, and will be detailed here later TODO.
Deep Space Tracking Radio Data
This functionality is still under construction
Radio tracking data from planetary spacecraft (Doppler, range, astrometry) collected by NASA’s Deep Space Network (DSN) or ESA’s ESTRACK network is disseminated through a number of channels, most notably the PDS geosciences Node, in a varity of data formats.
At the moment, Tudat is set up to read the Orbit Data File (ODF) files, documented here. These are binary files that Tudat can ‘unpack’ and put the contents into Tudat-compatible data structures. Since the contents of the radio science data are significantly more complicated than (for instance) optical astrometric data, the loading of the files is done in several steps:
Each ODF file is loaded into a single
OdfRawFileContentsobject. In this step, the contents of the binary file are loaded and put into basic C++/Python data types
The list of
OdfRawFileContentsobjects are processed, the relevant data combined and data structures set up, resulting in a set of
ProcessedOdfFileContentsobjects (each holding all data for a given link ends and observable type):
Ramp tables per ground station are created from the combination of all ODF files
All observations of a given observable type and link ends from all ODF files are merged into a single object holding the observables and relevant metadata
All observation times are converted to TDB
The properties of the ground stations (ramp tables) are taken from the
ProcessedOdfFileContentsobject and set in the environment using the
ProcessedOdfFileContentsto an object of type
ObservationCollection, which can be used in the estimation
To further use the
Pseudo-observations from External Ephemerides
Using some external source (for instance: Spice kernels) to compute/extract position observables (e.g. using the 3-dimensional Cartesian position of a body at an epoch as an ‘observable’), and then fitting these observations to a dynamical model in Tudat can be very useful. In particular, such a procedure allows you to quantify exactly how closely the dynamical model settings used in Tudat can recreate the published orbit. Using such Cartesian positions from an external data source is sometimes termed using ‘pseudo-observations’.
The source of the Cartesian positions is up to the user, but typical sources are:
Body positions from Spice kernels. NOTE: Spice kernels with spacecraft orbits for a large number of planetary missions can be found
Body positions from JPL Horizons
TLEs propagated in time using an SGP4 propagator, and rotated to an inertial frame
SP3c files containing tabulated state histories, typically for Earth-orbiting spacecraft
The Galilean moon state estimatione example on this page gives a good examples of the full procedure that can be used for this, where the states are (in this case) extracted from Spice kernels.
In Tudat Cartesian position (pseudo-)observations are processed using the
observation model. In addition to creating the
manually from external data, we provide a function of convenience to generate such pseudo-observations, using the following procedure:
Create the body for which the pseudo-observations are to be generated in your environment, using the
numerical_simulation.environment_setup.ephemeristudatpy module. Note that the
tabulated_from_existing()option can be used to turn any ephemeris settings into tabulated ephemeris settings (which is required if using the same bodies in the estimation).
Generate relative position observations (and associated observation model settings) using the
The latter function provides both the observations (as an
and a list of
ObservationModelSettings to be used
for simulating the observables. The combination of these two can be used directly for the subsequent steps
of defining estimation settings and performing the estimation.